Content uploaded by Wiebe E. Bijker
Author content
All content in this area was uploaded by Wiebe E. Bijker on Sep 27, 2018
Content may be subject to copyright.
This article was published in an Elsevier journal. The attached copy
is furnished to the author for non-commercial research and
education use, including for instruction at the author’s institution,
sharing with colleagues and providing to institution administration.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elsevier’s archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/copyright
Author's personal copy
Research Policy 36 (2007) 1088–1099
Techno therapy or nurtured niches? Technology studies
and the evaluation of radical innovations
Anique Hommels a,∗, Peter Petersb,1, Wiebe E. Bijkerc,2
aUniversity of Maastricht, Faculty of Arts and Social Sciences, Department of Technology & Society Studies,
PO Box 616, 6200 MD Maastricht, The Netherlands
bFaculty of Arts and Social Sciences, Department of Philosophy, PO Box 616, 6200 MD Maastricht, The Netherlands
cFaculty of Arts and Social Sciences, Department of Technology & Society Studies, PO Box 616, 6200 MD Maastricht, The Netherlands
Received 7 February 2006; received in revised form 19 March 2007; accepted 4 April 2007
Available online 23 May 2007
Abstract
This article contributes to recent discussions in technology studies about applying insights from technology studies to policy
decisions about the development and management of technological innovations. It does so by examining two approaches that can
be used by policy makers to manage radical technological innovations in mobility and transportation: strategic niche management
(SNM) and the PROTEE approach. The SNM approach uses protective ‘niches’ to develop radical innovations, whereas the PROTEE
method is grounded in the assumption that technological innovations have a better chance of success if made “vulnerable” by
subjecting them to risks and oppositions from the outset. Both SNM and PROTEE have, so far, been applied to retrospective case
studies. This paper examines their potential effectiveness in the monitoring of real time innovation projects by comparing their
conceptualizations of ‘learning’ and ‘experimenting’. It argues that the two approaches can draw upon each other to achieve a more
refined conceptualization of learning and experimenting and in dealing with the problem of change and obduracy in the development
of innovation projects.
© 2007 Elsevier B.V. All rights reserved.
Keywords: Strategic niche management; PROTEE; Technology studies; Learning; Experimenting; Innovation policy
1. Introduction
Explaining the success or failure of new technologies,
practices, artifacts, and regimes is a central concern of
technology studies. Over the past 20 years STS scholars
have studied technological innovation from a variety of
∗Corresponding author. Tel.: +31 43 3883483; fax: +31 43 3884905.
E-mail addresses: a.hommels@tss.unimaas.nl (A. Hommels),
p.peters@philosophy.unimaas.nl (P. Peters), w.bijker@tss.unimaas.nl
(W.E. Bijker).
1Tel.: +31 43 3883453.
2Tel.: +31 43 3883476.
perspectives. Their concern has been largely to exam-
ine the technical, economic, social, cultural and political
contexts of innovations. The question addressed in this
article is if and how knowledge about technological
innovation can be made relevant to communities and
organizations interested in developing policies for pro-
moting innovation. In this paper we argue that it is crucial
to know how knowledge about successful technological
innovations can be used to determine the potential suc-
cess or failure of the development of new technological
innovations. A second related question we address is how
these insights can be effectively translated into tools for
technology policy making.
0048-7333/$ – see front matter © 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.respol.2007.04.002
Author's personal copy
A. Hommels et al. / Research Policy 36 (2007) 1088–1099 1089
In the 1990s, the policy relevance of technology
studies became an explicit issue on the STS research
agenda (Rip et al., 1995; Sørensen and Williams, 2002).
Several attempts were made to turn insights garnered
from STS studies into instruments for monitoring inno-
vation processes. Several approaches were developed
including strategic niche management (SNM), bounded
sociotechnical experiments (BSTE), constructive tech-
nology assessment (CTA), the multi-level perspective,
the MIA approach and PROTEE. We will examine two
of these approaches: strategic niche management and
PROTEE.
Strategic niche management was developed by schol-
ars at the universities of Maastricht, Twente and
Eindhoven during the 1990s. It is a policy tool aimed
at concrete intervention in innovation processes. The
PROTEE approach was developed in a project under
the EU fourth framework programme, in a collaboration
between the University of Maastricht, Brunel University
(UK), the center for the sociology of innovation (CSI)
in Paris, and a number of industrial partners active in
developing new intermodal systems for freight trans-
port (Technicatome, Krupp, ZIV, Mondragon systemas).
PROTEE became further refined in the SOCROBUST
methodology (Jolivet et al., 2003) and in the MIA
approach (Maatschappelijke inbedding analyse: societal
embedding analysis) (Kets and Mourik, 2003).3
Innovations are traditionally divided into two types:
radical and incremental. Incremental innovations are
considered those which result from the development or
refinement of existing technologies. These tend to be rel-
atively easy to manage. In contrast, radical innovations
prove much more uncertain and risky and are therefore
more difficult to manage. Because radical innovations
are at the core of many research programmes in the pub-
lic sector and their financial consequences can be huge,
it is important to develop innovation management tools
for them. Both SNM and PROTEE attempt to take up
this challenge.
Strategic niche management (SNM) is defined as
the “creation, development and controlled break-down
of test-beds (experiments, demonstration projects) for
promising new technologies and concepts with the aim
of learning about the desirability (for example, in terms
of sustainability) and enhancing the rate of diffusion of
the new technology” (Weber et al., 1999 p. 9). The SNM
approach consists of a number of overlapping phases
(Weber et al., 1999). It starts by identifying a promis-
3Kets and Mourik (2003) tried to link the MIA-approach and tran-
sition management.
ing new technology or concept and by thinking about
the implications, advantages or disadvantages in broad
terms of the new technology. Second, an experiment has
to be designed. The new technology will be tested in
a controlled experimental set up in which the initiators
of the experiment decide about the actors that must be
involved in the experiment, and the protective measures
that must be taken in order to ensure its development.
The third stage consists of implementing the experiment.
The SNM approach identifies this stage as the most dif-
ficult (Weber et al., 1999) because it is at this point that
an experimental innovation encounters problems. The
problems which can bar successful implementation may
be technological as well as economic, social or insti-
tutional. The aim of this stage is to learn what kind
of problems arise during an innovation’s implementa-
tion so that technologies, social practices and mobility
assumptions can be adjusted. During the fourth phase,
the innovation process is scaled up from a single exper-
iment to a niche, and is then integrated with relevant
activities going on elsewhere. The fifth and final phase
is concerned with the evaluation of the level of pro-
tection of the niche. The protection is usually removed
after the experiment is completed, but sometimes it is
removed during an experiment. The decision whether
to continue or cut an experiment is made during this
stage.
PROTEE is designed as a real-time management tool
for evaluating the quality of ongoing technological inno-
vation projects. Its approach consists of an iterative
process in which an innovator and an evaluator engage
with each other. The interaction between innovator and
evaluator has the form of a socio techno therapy. The
innovator gives a description of a project’s risks based
on the evaluator’s PROTEE-checklist. The checklist con-
sists of the four classes of indicators.4The evaluator must
elicit a risky description of the project using these indi-
cators as questions in an interview. Each indicator can
be assessed on a scale of 1–5. After a couple of weeks or
months the ‘therapeutic’ session is repeated. The innova-
tor gives another description of the present state of affairs
in the innovation project. The evaluator then compares
4The four classes are called realism, strategy, falsifiability and
innovativeness. Examples of indicators (within these four classes) are:
how much opposition do the innovators take into account? How diverse
and independent are the trials to test the various aspects of the project?
Is the project learning from its environment? For more details, we refer
to the final report of the PROTEE project (PROTEE final report (EU
4th framework), 1999). See http://europa.eu.int/comm/transport/extra/
web/downloadfunction.cfm?docname=200310/protee.pdf&apptype=
application/pdf.
Author's personal copy
1090 A. Hommels et al. / Research Policy 36 (2007) 1088–1099
the two descriptions, assigns a grade and by this process
assesses the quality of the learning curve.
Both SNM and PROTEE have, so far, been applied
to retrospective case studies. STS scholars have used
both PROTEE and SNM to study innovation processes
in transport systems.5PROTEE has been used in studies
of innovations in freight transport, such as the build-
ing of a fully automated container terminal (see below),
the construction of new freight handling systems and
the implementation of a new transport and freight cen-
tre. SNM has been used for studying personal mobility
technologies, in particularly, electric vehicles, car shar-
ing systems, sustainable public transport and traffic
management systems. It is interesting to see that most
analyses using PROTEE and SNM are done ex post,
whereas both are usually presented as real-time moni-
toring and in the case of SNM, forward looking. Both
approaches put strong emphasis on experimenting and
learning processes.6Although experimenting and learn-
ing are interpreted differently in the two approaches, both
approaches emphasize that the primary goal of experi-
menting in innovation processes is to increase the ability
to learn about these processes. Learning can apply to
learning about the technical aspects of an innovation,
but also to learning about its social environment, the user
context, relevant government policies etc.
Despite sharing a focus on learning and experiment-
ing, the two approaches differ in crucial ways. In the
first place, the two models differ in their objectives:
SNM wants to achieve a more sustainable society by
stimulating sustainable technologies, whereas PROTEE
aims at optimizing learning processes of, in theory, all
kinds of technology. Secondly, there are important differ-
ences in the theoretical underpinnings of each approach.
SNM draws on evolutionary economic approaches to
innovation, constructive technology assessment, and
sociohistorical studies of past transformations. PROTEE
draws on general STS studies of innovation processes
and uses concepts from actor network theory and SCOT.
Recently, some attempts have been made to merge the
fields of STS and evolutionary economics.7Interest-
ingly, Russell and Williams (2002) see SNM as an
5Work in SNM has also been carried out in the areas of energy and
waste water. PROTEE is claimed to have a wider applicability.
6The same applies to BSTE. BSTE also views experiments as pos-
sibilities to learn and is less aimed at intervening in the innovation
process than SNM and PROTEE (see Brown et al., 2003).
7See for instance Geels (2004). According to Russell and Williams
(2002) evolutionary economics (EE) and the social shaping of tech-
nology approach are largely compatible: “their preoccupations overlap
strongly and their findings are often consistent” (p. 44). There are,
however, important points on which social shaping theorists (SST)
example of “the fruitful interaction of sociological and
evolutionary economic ideas” (Russell and Williams,
2002 p. 45).
In this article we examine these two models in more
detail. We will begin by describing the key elements
of the SNM and PROTEE approaches and the theoreti-
cal assumptions underlying them. The observation that
the political goals of the two models differ crucially
results, for instance, in different criteria for the suc-
cess of the interventions made in the innovation process.
Moreover, we claim that the theoretical assumptions
underlying them are based on different views of the
structure of experiment. In the SNM approach, exper-
imenting is understood as creating a controlled space, a
‘niche’, in which most contingencies can be ruled out.
In contrast, in the PROTEE approach, it is precisely the
continuous confrontation between innovation and con-
tingency that constitutes the experiment and ultimately
determines an innovation’s success or failure. Since both
approaches underscore the importance of learning pro-
cesses for the realization of technological innovations,
we will also evaluate the potential of each for improving
the quality of the learning processes entailed in techno-
logical innovation trajectories. Finally, we will discuss
the implications of using models like SNM and PROTEE
for policy development.
2. Developing innovation in protected spaces:
strategic niche management
SNM is geared to deal with innovations that are likely
to contribute to a regime shift. A typical example of
such a shift is the move from the existing energy sys-
tem based on non-renewable (fossil) energy sources to
a more sustainable energy regime based on biomass.8
By focusing on transitions to a more sustainable soci-
ety, the SNM approach has a clear political goal. In
the SNM approach, an innovation process is considered
remain critical of evolutionary economics: EE focuses too much on
the macro/meso level and fails to acknowledge and explain diversity
and contingency in innovation processes (p. 44). It tends to overlook
the detailed dynamics and outcomes of a particular innovation process.
Furthermore, EE tends to treat local learning processes as black boxes.
Russell and Williams argue that the SST approach is better equipped
to address the detailed mechanisms involved in innovation. However,
“we should not expect a complete merging of the two bodies of theory:
their agendas and concerns are different, and they have strengths at
different levels of generalization” (Russell and Williams, 2002 p. 45).
8See Rob Raven’s PhD thesis (2005). Another example is the regime
shift from the car system based on the electric combustion engine
to a car regime based on more less polluting energy sources such as
electricity.
Author's personal copy
A. Hommels et al. / Research Policy 36 (2007) 1088–1099 1091
successful when it results in what is called a ‘regime
shift’. Since the aim of SNM is primarily to achieve
regime shifts which contribute to a more sustainable
society (Hoogma et al., 2002), the implementation of
new, more sustainable technologies is crucial. In the
SNM approach, new technologies are ideally developed
in protected spaces or niches. However, Hoogma et al.
concluded in a recent study that many of these innova-
tions are never adopted. It turns out to be very difficult to
implement technological innovations outside the niche
in which they were developed, which is what led Hoogma
et al. to the disappointing conclusion that “we were cer-
tainly too optimistic about the potential of SNM as a tool
for transition” (Hoogma et al., 2002 p. 195).
The success of early niche development is measured
by two criteria according to Hoogma et al. (p. 28):
the quality of the learning processes and the quality
of institutional embedding. Hoogma et al. propose a
broad concept of learning that involves learning about
the technologies, but also learning about users, soci-
etal and environmental impacts and government policy.
Institutional embedding means that in the process of
development from niche to regime, a kind of network
is built to support the innovation. This network includes
complementary technologies and infrastructures, but
also a number of supporting actors (e.g. users), guided
by a shared set of expectations.
SNM is based on evolutionary economics, which
strongly influences its model of change. Its model of
change is based on the notion of path dependence, the
classic example of which is the QWERTY-keyboard.9
The change in technological regimes is explained by
referring to actors who have a stake in searching for
incremental solutions that can be achieved by improv-
ing the dominant technology. It makes economic sense
for them to stick with the dominant technology because
there are many investments (in terms of money but also
users skills) made in that technology (Geels, 2002). It
is interesting to note that issues of path dependence are
temporarily set aside or put between brackets in the SNM
approach: “In the niche model, lock-in and path depen-
dency assumptions are relaxed. Various technological
options can co-exist over a long period, precisely because
of the existence of niches requiring other functionalities”
(Hoogma et al., 2002: 26).
9The QWERTY-keyboard was developed in the era of mechanical
typewriters as a solution to the recurrent interference of adjoining keys.
In the period of personal computers and electronic keyboards, however,
this arrangement no longer seemed necessary, and alternative designs
were made. None of these, however, became widely accepted.
This line of reasoning becomes clearer when we
relate the SNM approach to the use of path dependence
and lock-in in the so-called “multi-level perspective”.
The “multi-level perspective” distinguishes three levels
in technological development, diffusion and implemen-
tation of innovations (Geels, 2004). The micro-level
is the level of niches and the meso-level is the level
of “technological regimes”.10 According to Rip and
Kemp,
“a technological regime is the rule-set or grammar
embedded in a complex of engineering practices, pro-
duction process technologies, product characteristics,
skills and procedures, ways of handling relevant arti-
facts and persons, ways of defining problems—all of
them embedded in institutions and infrastructures”
(Rip and Kemp, 1998 p. 338).
In contrast to the niche level where the rules are not
yet stable or clear, in regimes rules have become stable
and have more structuring effects on the activities per-
formed in the regime (Geels, 2004). At this level, change
becomes more difficult to achieve as the regime limits the
development of alternatives that do not fit into it. More-
over, because regimes are structures, individual actors
have difficulties changing the rules that form its basis
(Rip and Kemp, 1998).
The macro-level is the so-called “sociotechnical land-
scape”. The metaphor of landscape puts stress on the
obduracy of any given society’s material and immaterial
makeup. Landscapes consist of material environments,
shared cultural beliefs, symbols and values. They are
difficult to change because “landscapes are beyond the
direct influence of actors, and cannot be changed at
will” (Geels, 2004 p. 913). It is interesting that the
more we move to the macro-level, the greater the
acknowledgement of constraining factors and resistance
to change is.11 This means that the potential for flex-
ibility and change are linked to the level of niches,
whereas embeddedness and resistance to change enter
the model at the meso-level of regimes. This implies
that at the level of niches, obduracy of sociotechnol-
ogy is hardly taken into account. This explains the
optimism about the potential for change in much of
10 The concept of technological regime in the multi-level perspec-
tive is a broadening of Nelson and Winter’s concept: “Regimes are a
broader, socially embedded version of technological paradigms” (Rip
and Kemp, 1998: 388).
11 This can be restated by referring to Thomas Misa’s argument that
the more we move to the macro-level of analysis, the more we are
inclined to take a technological determinist point of view (Misa, 1988).
Author's personal copy
1092 A. Hommels et al. / Research Policy 36 (2007) 1088–1099
the SNM literature.12 Part of the constraining mech-
anisms that play a role in the ‘selection environment’
are at least temporarily ignored. This is different in the
PROTEE approach where no distinctions between lev-
els are made. Here, PROTEE’s roots in ANT become
clear: distinctions are the result of learning and inno-
vation processes but cannot be distinguished a priori.
It is interesting that in recent SNM-literature, a too
strong distinction between these levels (e.g. between
niche and regime) has been criticized (see e.g. Raven,
2005).
3. Innovation management as a socio techno
therapy: the PROTEE approach
The PROTEE model is designed for innovation
projects of which the outcome is uncertain and unclear.
Managing these kinds of technological innovations is
difficult because the uncertainties surrounding them
prevent the use of conventional management tools to
measure risk and the probability of success. This raises
the question of whether such projects can be managed
at all and, if so, how. PROTEE is limited to procedu-
ral indicators only. The key feature of PROTEE is that
projects are not evaluated according to costs, feasibility,
social acceptability, plausibility, coherence etc., but only
through the quality of the procedure used to learn more
about costs, feasibility, social acceptability etc.
The starting point of the PROTEE approach is a
‘learning pact’ between the innovator and a PROTEE-
equipped evaluator. The evaluator uses a checklist that
contains a number of questions that are meant to make
the innovator aware of mistakes that are typically made
in developing innovations. Drawing on STS case studies,
the PROTEE approach has identified four basic reasons
for the failure of technological innovations:
12 It is interesting to note that SNM and “multi-level approach” stud-
ies appear to diverge on the optimism they express in relation to the
possibility of achieving change. On the one hand, Geels stresses that
as long as there is a stable regime, radical changes at the level of
niches cannot be expected. Transitions come about when dynamics at
the three levels link up and reinforce each other (Geels, 2004). On the
other hand, Kemp et al. (2001) quite optimistically focus on the oppor-
tunities the regimes and landscapes create for sociotechnical change.
This difference in expectations can perhaps be explained by the con-
text of their research. Geels studies historical cases, whereas those of
Kemp, Rip and Schot are more policy-oriented. In a policy context –
aimed at transforming our society into a more sustainable one – a more
optimistic view of the possibilities to achieve change definitely sounds
more productive.
(1) Lack of realism: This class criticizes technological
innovations that are seen as ballistic: passing from
non-existence into existence without taking the con-
text into account. This class makes the innovator
aware of different degrees of uncertainty in his or her
project, which, if recognized, can lead to alternative
paths.
(2) Lack of strategy: This class criticizes the inability
of innovators to take into account various forms of
resistance to a project. This class urges an innovator
to learn about (potential) opposition to a project and
anti-programs.13 Anti-programs are human or non-
human entities whose behaviour may put the success
of the project at risk.
(3) Lack of falsifiability: This category criticizes the
practice of innovators to avoid circumstances in
which their accounts of the project could be reliably
falsified. The trials to test a project’s development
must be representative and critical. This class of
indicators makes the innovator aware of ways to test
critically the evolution of a project.
(4) Lack of innovativeness: This class criticizes the prac-
tice of stopping innovation projects because they
are too risky (hopeful monsters) and of fostering
projects that are not worth continuing (hopeless
pets). This category makes it possible to distinguish
between hopeful monsters and hopeless pets.
A relationship between an innovator and an evalua-
tor is established in the process of evaluating a project
according to these four classes of mistakes. The two
engage in a learning process to analyze and discuss the
project in PROTEE terms. The comparison of the indi-
cators over time makes it possible to establish the quality
of the learning curve. The evaluator makes sure that he or
she avoids the four classes of mistakes in technological
innovation in order to maximize the chance of navigating
the project through its development. An explicit indica-
tor is formulated against each mistake. To avoid the four
classes of mistakes, means that the learning process will
be optimal. In the end, innovators and evaluators have to
reach a common definition of the project to decide over
its prolongation or ending.
An important feature of PROTEE is that it attaches
a great deal of importance to processes of learning and
experimenting. PROTEE seeks to enable an innovator
and an evaluator to establish a ‘learning pact’ which
enables the quality of the learning process to be assessed.
The project is conceived as a learning process, aiming
13 On anti-programs, see Latour (1997).
Author's personal copy
A. Hommels et al. / Research Policy 36 (2007) 1088–1099 1093
at the progressive discovery of the context in which the
innovation becomes or will be embedded. It is assumed
that neither the innovator nor the evaluator begins in a
state of knowledge: they do not know how to realize
a technological innovation. That is why they have to
learn. Learning is applied to all relevant variables in an
innovation project: learning about clients, risks, costs,
social acceptance, etc. This implies that multiple learn-
ing curves can be distinguished in any given project, each
of which is related to the different variables that are in
play in an innovation process. The PROTEE approach
does not accept innovators who always give the same
description of a project, because doing so implies that the
learning curve has not begun. In this respect, PROTEE
differs from traditional policy tools in which the repeti-
tion of the same presentation is taken as an indicator of
the robustness of a project.
PROTEE defines a successful innovation trajectory
as a process in which learning is maximized. As a con-
sequence, we have to redefine success and failure: for
PROTEE, a failed project is one that does not learn. If
it is impossible for anyone involved in the project to
learn about its context, its opponents, its redefinition,
or the trials necessary to judge it, if the same smooth
description is always given, then the project should be
stopped. These starting points feed back on the defini-
tion of innovation in the PROTEE approach. PROTEE
characterizes innovation by taking risks under uncer-
tainty, with a lack of knowledge. If a less risky choice
is made, less can be learnt and thus PROTEE will be
less capable of improving the learning curve. PROTEE
not only focuses on the technical parts of projects but
also on innovations in usage, control, accounting, law, as
well as pieces of machinery or composition of existing
machinery (PROTEE final report (EU 4th framework),
1999).
In contrast to SNM, the PROTEE approach is geared
primarily towards maximizing the learning capacity of
an innovation project rather than the adoption of new
(sustainable) technology on the grounds that the better
the learning curve of a project is, the more chances it
has of survival in the real world. The way PROTEE
works in launching technological innovations differs
from SNM. Instead of initially protecting innovations
in niches, PROTEE deliberately focuses on the vulnera-
bilities of the innovation process: it stimulates innovators
to focus on those aspects in the innovation process that
are difficult, risky or likely to fail.
Having discussed the key features of both approaches,
we will now compare their strengths and weaknesses and
their (in)compatibilities in relation to two themes: (1)
the underlying notions of the structure of good experi-
ments (Section 4) and (2) the different ways in which
both models relate learning processes to sociotechnical
change (Section 5).
4. Experimenting: making innovations
vulnerable or protecting them?
One of the first theorizers of the innovation process,
Joseph Schumpeter already stressed the importance of
vulnerability for innovation (Schumpeter, 1942). He rec-
ognized that the fundamental instability of the capitalist
system always presented opportunities for entrepreneurs
to engage in innovative enterprises. In doing so, Schum-
peter identified vulnerability, seen previously only as
a state of instability and uncertainty, as a positive and
necessary prerequisite for innovation (Bijker, 2006).
Accepting Schumpeter’s claim as a precondition of inno-
vation, we argue that one of the key differences between
the two models is their conception of the relation between
an innovation and its context. In the SNM approach,
radical innovations are initially developed in protected
niches after which they are confronted with the ‘real
world outside’. The PROTEE approach takes the oppo-
site trajectory and begins the development process by
trying to make any given technological innovation vul-
nerable, with the aim of maximizing its learning capacity.
SNM advocates that innovations be developed in pro-
tected niches or ‘nurtured spaces’ (Hoogma et al., 2002).
A variety of policy instruments are then used to protect
them, e.g. government subsidies, partnerships, or policy
interventions. Niches form a nurtured space for the gen-
eration of radical innovations. Niches offer a basis for
their further development, for new development trajec-
tories, and for the transformation of regimes. Kemp et
al. (2001) stress that the SNM approach was originally
developed to increase the chances of a new technology’s
successful introduction. Later the approach became part
of a broader framework whose aim was to build new tech-
nological regimes and to create possibilities for regime
shifts (Kemp et al., 2001: 270).
However, according to Hoogma et al. (2002), one of
the issues that is still partly unresolved is how niche
protection should be organized and how the gradual
reduction of protection can be carried out without dis-
rupting an innovation’s development. Hoogma et al.
argue that forms of protection available locally may not
be enough. They claim that a more intensive form of
protection accompanied “by sponsors and accompany-
ing measures that change the overall frame conditions
for economic decision making” is necessary (Hoogma
et al., 2002: 202). They acknowledge however, that one
of the disadvantages of such intensive forms of protec-
Author's personal copy
1094 A. Hommels et al. / Research Policy 36 (2007) 1088–1099
tion is that niches can act as prisons (Hoogma et al., 2002:
171). This phenomenon helps to explain why it is so dif-
ficult for an innovation to become adopted once it has left
its protective niche. A number of empirical studies using
the SNM approach have confirmed this observation. It is
striking that the planned top down experiments described
by Hoogma et al. (2002) (the Bikeabout project, ASTI
and Praxit`
ele14) failed, whereas car sharing, as a newly
emerging experimental business, was the most success-
ful in developing outside the niche. Another example of
a retrospective SNM-experiment is the PIVCO electric
vehicle developed in Norway. The new vehicle received
a lot of protection from some of Norway’s major indus-
tries and the government. However, when the protection
was removed and the vehicle was tested outside its pro-
tective niche, the public did not accept it. Moreover, the
circumstances created in the niche were so typically Nor-
wegian that the concept could not be transferred to other
countries (Hoogma et al., 2002; Weber et al., 1999).
Another example is a project to develop electric vehi-
cles on the German island of R¨
ugen. In October 1992, a
testing and evaluation project of electric vehicles began
in R¨
ugen which lasted 4 years. Sixty EV’s equipped
with new battery systems were tested by more than a
hundred users. The project was subsidised by various
German states and municipalities. Hoogma et al. stress
that without a subsidy provided by the German federal
research ministry covering nearly 50% of the project’s
costs, it was unlikely that the car manufacturers would
have initiated the experiment. The results of the exper-
iment were disappointing. According to Hoogma et al.
“the contribution of the R¨
ugen project to the develop-
ment of the electric vehicle niche was very limited with
respect to both learning and institutional embedding”
(Hoogma et al., 2002 p. 71). Learning was limited to
technical issues, such as the testing of new types of bat-
teries. And although the environmental benefits of the
project (both in terms of lower noise levels and less
pollution) were clear, the manufacturers were not pre-
pared to arrange a follow-up to the project even though
it was recommended by the project manager. The num-
ber of EV’s on R¨
ugen in 1996 was a bit more than at the
start of the project in 1992, but it clearly did not con-
tribute to further niche development, as Hoogma et al.
claimed. Despite these disappointing results, SNM pro-
14 The Bikeabout project started in the early 1990s in Portsmouth, UK
as an experiment to set up a bicycle pool. The accessible sustainable
transport integration project in Camden, UK ran between 1994 and
1997 and involved the development and implementation of a number
of electric and natural gas powered minibuses. Praxit`
ele was a French
project to stimulate individualized public transport with EV’s.
ponents did not conclude that formal experiments have
less potential for SNM. They admitted, however, that it
is important to think about “the trade-offs between an
open setting (which facilitates learning) and tight con-
trol (which closely steers the experiment)” (Weber et al.,
1999 p. 37).
The difficulty experienced in SNM with the societal
embedding of innovations after niche conditions have
been removed can arguably be explained by the concep-
tual underpinning of the SNM approach itself. SNM is
based on an evolutionary model of technological change
in which variations are developed within niches and
selection takes place within a selection environment. The
approach thus assumes a rather clear distinction between
a technology and its context. Only in the third phase
of an SNM process (see above), is an innovation con-
fronted with its selection environment, and actors begin
to think about possible opposition to the project. After a
niche has been successfully established, the ‘real’ selec-
tion environment still has to be confronted, because the
niche only partially reflects, ‘reality’. In this respect, it
is salient that Hoogma et al. (2002) concluded that the
projects they studied were “overly self-contained”. This
can be the result of the experimental set-up in niches,
which forms the cornerstone of the SNM approach.
Whereas SNM rests on the assumption that technol-
ogy and its context can be (temporarily) separated the
PROTEE approach opposes this idea. The key innova-
tion of the PROTEE approach (and here it differs most
from other approaches to managing innovations) is that
it deliberately tries to make an innovation vulnerable by
confronting it as much and soon as possible with the
‘real world’. The means of achieving this are based on
eliciting a set of risky descriptions—descriptions, made
by the innovator, in which uncertainties, opposition, and
alternatives to the present development trajectory are
made explicit. Normally, policy makers want to make
sure that a project they fund will be described in the
‘smoothest’ possible way. Innovators will ‘by nature’
start with a persuasive description of their projects. In
the PROTEE approach, it is a contradiction of terms to
say that a technological development is an innovation
and to give a description of it in which all the issues that
may jeopardize the future success of the project, are left
out. Therefore, acknowledging the potential vulnerabil-
ities of any given project is a very important feature of
PROTEE, and that encapsulates exactly what we mean
by a risky description of the project. By “vulnerability”
we do not mean only that a project should be open to
criticism, but also that any criticisms of it needs to make
vulnerable in advance key features of the project in order
to anticipate future critical developments.
Author's personal copy
A. Hommels et al. / Research Policy 36 (2007) 1088–1099 1095
An example of a technological innovation that has
been analyzed (retrospectively) using the PROTEE
approach is the automation of the Delta-Sealand termi-
nal in Rotterdam’s harbour between 1982 and 1993.15
Berghoff et al. have argued that the Delta-Sealand ter-
minal represents a fundamental innovation in container
transhipment (Berghoff et al., 1993). The terminal’s
automation entailed four innovations: (1) transporting
containers with automated guided vehicles; (2) stacking
containers with automated stacking cranes; (3) a comput-
erized operating system; and (4) a number of changes in
the organization of labor.
According to some PROTEE indicators, the inno-
vator, Europe combined terminals (ECT) – one of the
world’s largest stevedores – learnt a lot during the
project’s first stage, which lasted from 1982 to 1986.
ECT wanted to build an automated terminal and was
looking for customers. It broadened its network and
thought strategically about the various actors that would
be needed to build the terminal. When Sealand entered
the scene as a potential customer, ECT was flexible
enough to offer four different terminal designs. In other
aspects, however, the learning process was less success-
ful: it is unclear how much attention ECT paid to possible
anti-programs. The risk of sealand withdrawing from
the negotiations was clearly identified, but other risks
were not considered. By PROTEE criteria, this neglect
of potential risks is considered a bad sign.
In a later stage, which took place between 1987 and
1992, less was learnt in this project: there was a clear
choice for one terminal design, and the layout of the
project as a whole had become somewhat inflexible. The
technical specifications of the terminal, agreed upon by
sealand, had to be met at any cost. During this stage of
the project, the PROTEE approach was less useful as a
tool for managing technological development because
the project had more or less stabilized, the uncertainties
had diminished and less was learnt.
One major weakness in the PROTEE approach is that
it remains unclear how to decide between the stabiliza-
tion of a project when an innovation proves successful or
to stop an innovation’s development because the innova-
tors do not learn anymore. When does the relevance of the
PROTEE approach end and when does it become more
useful to use other technological management tools? At a
certain moment, the uncertainty in a project diminishes,
the degree of innovativeness of a project declines, and
less is learnt. Then PROTEE is not useful any longer,
15 For a detailed analysis of the process of designing and building the
delta-sealand terminal in Rotterdam, see Bijker et al. (1999).
but it is difficult to assess precisely when this moment
occurs. This is perhaps comparable to the difficulty in
SNM to determine when niche conditions can be safely
removed. We think, however, that the risk of ‘a niche
becoming a prison’ can be solved by confronting the
innovation with ‘the real world outside’ in an earlier
stage. Interestingly, it can be argued that the experiments
described in Hoogma et al. can be considered successful
according to PROTEE criteria if one looks at how much
has been learnt during some of the experiments.
5. Learning: conceptualizing processes of
sociotechnical change
It is clear that SNM, like PROTEE, aims at achieving
an enhanced reflexivity among actors involved in innova-
tion processes. STS scholars have argued that when the
SNM approach is used, actors will acquire a better under-
standing of the nature of technological change and the
dilemmas that might arise in managing it. Learning is a
key concept in the SNM approach as well. In SNM, learn-
ing is broadly conceived as learning about technical spec-
ifications, the user context, societal and environmental
impacts, industrial developments and government policy
and regulatory frameworks. Hoogma et al. (2002) distin-
guished between first order and second order learning in
the SNM approach. First order learning means that, in a
niche, actors learn about how to improve the design of
a technological innovation, which features of its design
are acceptable for users and ways of creating a set of pol-
icy incentives that will facilitate its adoption. However,
second order learning is needed for the establishment of
a regime shift on the basis of niche development. In sec-
ond order learning, conceptions about technology, users,
demands and regulations are not tested, but questioned
and explored. This is called “co-evolutionary learning”
(Hoogma et al., 2002: 29). Successful niche develop-
ment consists of first order learning on a whole array of
aspects coupled with second order learning.16
The process of learning is much more operationalized
in PROTEE than in SNM: learning indicators are defined
and the whole technological innovation management
instrument is designed to be able to assess the quality of
the learning curve. At the same time, both approaches
assume that optimal learning processes increase the
chances of an innovation’s successful implementation.
However, the definition of a successful innovation pro-
cess differs in PROTEE and SNM. For SNM, success
16 Hoogma et al. (2002) conclude, however, that first order learning
is far more predominant in the experiments they studied.
Author's personal copy
1096 A. Hommels et al. / Research Policy 36 (2007) 1088–1099
consists of the creation of niches for an innovation, an
innovation’s development and ultimate diffusion. For
PROTEE success is defined as the achievement of an
optimal learning process.
How does learning lead to sociotechnical change
in both approaches? As mentioned above, the SNM
approach is heavily based on the notion of path depen-
dence. Social constructivists describe “trajectories” or
“paths” as being actively constructed or destructed,
instead of being given by nature (Garud and Karnøe,
2001). This approach seems to have been adopted by
SNM theorists as well. According to Kemp et al.,
“SNM is a method for constructing paths” (Kemp et al.,
2001: 270). However, one of the problems of models of
path dependency is that they usually ignore the role of
agency.17 Crude notions of path dependence and trajec-
tories as developing according to an internal, “natural”
logic have been criticized by STS scholars, who empha-
size the contingent and fluid character of technological
development. The sociologist of technology Trevor
Pinch has correctly pointed out the “a-symmetric” use of
the notion of path dependence by economists, who only
invoke history when addressing “inferior” technologies.
In contrast, Pinch proposes a more constructivist vari-
ant that focuses both on the paths taken and alternative
paths not taken (Pinch, 2001). It is precisely because the
SNM approach starts with a clear conception of the end-
result of an innovation (that is: a sustainable technology)
and temporarily shields this conception in niches from
the ‘real world’, that the possibility of alternative paths
which arise as a result of learning processes are ruled out.
Whereas in the SNM approach, an innovation seems
to be taken as more or less “ready made”, the PRO-
TEE approach considers any innovation as constantly
“in the making”. PROTEE clearly starts from differ-
ent assumptions about the relation between learning and
sociotechnical change. Because PROTEE begins from
an interaction between an innovator and an evaluator, the
actor perspective18 is much more visible in this approach
than in SNM. PROTEE, which is based on sociological
and anthropological approaches to understanding tech-
17 For a review of critiques of the notion of path dependence see
MacKenzie and Wajcman (1999). The critique comprises two claims:
(1) it is not clear, but often disputed what the “best” technology is
(maybe the alternatives to QWERTY are not superior); (2) if a tech-
nology has an existing alternative that is seen as better by many, it
is unlikely that the “inferior” technology will survive, since there are
many strategies (e.g. subsidizing or offering it belowcosts) to overcome
the lock-in context of the “inferior” technology.
18 On the importance of an actor perspective in designing mobility
projects, see Peters (2006).Smith et al. (2005) also argued that ‘agency’
should play a more important role in regime transformation models.
nological innovation such as SCOT and ANT, focuses
on the work that the actors must do to achieve change.
In contrast to SNM, learning in the PROTEE approach
means opening up the possibility of alternative paths.
More specifically, in the PROTEE approach, a class
of indicators is directed specifically to the problem of
obduracy and the possibility of developing alternative
“paths”. This class of indicators (called strategy)
focuses on the degree to which innovators take possible
opposition to their project into account: “the innovator
should be able to describe the project by populating the
world with as many opponents (human and non-human)
as possible. The first thing to test is the absence or
presence of anti-programs, that is, of entities etc.,
whose behaviour may jeopardize the project” (PROTEE
manual, test version: 13). Secondly, the innovator must
be able to describe his or her project from the point of
view of its opponents. The innovator should be able to
give “good reasons” why they might oppose the project.
Third, the innovator must have a clear view of which
aspects of the project are negotiable with opponents and
which aspects that are not. Finally, the innovator must
make sure that his project is flexible enough to incorpo-
rate opposition in the project’s design. The project must
be able, in other words, to “absorb” as much opposition
as possible (PROTEE manual, test version).
The Delta-Sealand terminal case illustrates the PRO-
TEE approach. In the beginning of the project some
anti-programs were identified. For example, the main
customer of the terminal, sealand, would drop out of the
negotiation process. In the early stages of the terminal’s
development (1982–1986), the project was learning from
its environment in the sense that the terminal design was
put forward as a way to solve the problem of social unrest
in the Rotterdam harbor and the perceived expansion of
container transhipment in the near future. In a later stage,
organizational anti-programs were identified. In opera-
tional tests of the new terminal, it turned out that some
of the crane drivers did not operate the cranes in the way
that software engineers had expected. PROTEE’s aim is
to make innovators think about such obstacles from the
start of their projects.
The point where the PROTEE and SNM approaches
can learn from one another becomes clear when we
describe the results of the R¨
ugen case. Importantly, the
actors who challenged the conventional wisdom about
car manufacturing, small car manufacturers, critical sci-
entists or environmentalists, were excluded from the
project. Users were involved only in the testing phase
and excluded from the design and setting-up phases of
the project. Hoogma et al. (2002) suggest that if such
opposition to the project would have been taken more
Author's personal copy
A. Hommels et al. / Research Policy 36 (2007) 1088–1099 1097
Table 1
Comparison of the two models of innovation management
PROTEE SNM
Disciplinary foundations STS, SCOT, ANT STS, evolutionary economics
Political objective No explicit political objective Political objective is aimed at realizing a
specific path (towards sustainability)
Criteria of success Optimal learning process (leading to a
surviving technology)
More sustainable technology (leading to a
more sustainable society)
Experimental method Making innovations vulnerable (by socio
techno therapy)
Initial protection using ‘niches’, gradual
removing of niche protection
Key mechanism in innovation processes Integration of innovation and context from
inception, role of obduracy, embeddedness
Distinction between variation and selection
environment, path dependence
Related tools Socrobust, MIA CTA, alternative technology (AT), multi-level
approach
seriously, the niche development process might have
been more successful.19 If, indeed, PROTEE would have
been applied in the R¨
ugen case from its inception, the
innovators might have been more aware of the critical
(f)actors hampering the project.
In sum, because of its theoretical roots in evolution-
ary economics, SNM’s model of technological change
is closely linked to the notion of path dependency. PRO-
TEE, having its roots in SCOT and ANT, is based on
notions as obduracy, actor perspectives and embedded-
ness. Because in the SNM and “multi-level” approaches
the role of constraints to innovation processes is lim-
ited to the macro- and meso-levels of development, we
have argued that PROTEE offers more explicit ways of
dealing with opposition to a project and takes the obdu-
racy of technology more seriously. On the one hand,
it is understandable that path dependence is so central
in SNM, considering SNM’s political goal. Path depen-
dence (e.g. of a sustainable technology trajectory) can be
very positive in innovation processes. A clear advantage
of the SNM approach in this respect is that, by focus-
ing on path dependencies and institutional embedding,
this model has a clearer idea of how to stabilize an (desir-
able) innovation. In the PROTEE approach it is less clear
how and when to reach ‘closure’ in the development of
a radical new technology. Furthermore, PROTEE might
be subject to the critique that an actor oriented approach
assumes that it is always clear which actors are involved,
what their roles are, and that the whole process can be
shaped by the actors at will and without much attention
19 Hoogma et al. describe four EV experiments. Apart from the
PIVCO and the R¨
ugen cases, they studied a project in La Rochelle,
France. In this case, the EV sales were ultimately disappointing because
electric vehicles for private use could not exist without special protec-
tion. Another EV-experiment in Mendrisio, Switzerland contributed
mostly to niche development because it was able to create a positive
image for EVs.
for structural factors. Although the PROTEE model is
not intended to be voluntaristic, a clearer focus on insti-
tutional embedding, as in SNM, can clarify this issue.
6. Conclusion: policy implications of technology
studies
This article examined the SNM and PROTEE
approaches to managing technological innovation by
describing their key elements and the theoretical assump-
tions underlying them. An overview of the main
differences between PROTEE and SNM discussed in this
article is provided in Table 1.
We acknowledge that the two approaches must be
tested in real time innovation projects to be able to eval-
uate them properly. The aim of this article is not to state
which of the two models is ‘best’, but to discuss their
strengths and weaknesses. Both approaches have a very
rich understanding of learning in innovation processes, in
which a great variety of elements can be subject to learn-
ing: technical aspects, but also user aspects, legal issues,
or financial facets. An interesting feature of SNM is that it
makes a distinction between first order and second order
learning. However, learning is much more operational-
ized in PROTEE. The interactions between evaluator
and innovator in PROTEE are aimed at establishing the
quality of the learning curve. It is less clear how the
quality of the learning process is analyzed in the SNM
approach. Moreover, in PROTEE, learning has a clear
temporal dimension: the quality of the learning process
can change over time. In SNM, this aspect of the learning
process is unclear. Furthermore, ‘learning’ in the PRO-
TEE model acknowledges the important role of barriers
to sociotechnical change and obduracy of sociotechnol-
ogy. Underestimating the factors or actors that constrain
radical change makes the SNM model perhaps less capa-
ble of dealing with the problem of obduracy. Moreover,
Author's personal copy
1098 A. Hommels et al. / Research Policy 36 (2007) 1088–1099
a conception of obduracy that is mainly based on the
notion of path dependence (as in SNM) keeps innovators
unaware of other mechanisms involved in the construc-
tion of obduracy.20
In terms of their policy relevance, both approaches
have something different to offer.21 The SNM approach
is related to a clear political agenda in the sense that
striving for a more sustainable society is its main goal.
This can only be achieved by starting radical innova-
tions out in protected niches. PROTEE, on the other
hand, does not have such a political goal. This lack of
normative engagement is a consequence of its underly-
ing theoretical perspective in which a desired outcome
is, deliberately, not stated beforehand. In theory, PRO-
TEE can be thus used to monitor all kinds of radical
innovations. Whether the PROTEE-innovation process
contributes to a more sustainable society can only be
assessed retrospectively and is not the starting point.
The policy relevance of PROTEE is that it offers
insights into the learning capacity of innovation projects
and tentatively helps to determine when a project should
be stopped or continued. However, PROTEE is disap-
pointing in not yet being specific and clear enough as
to when this moment occurs. Here PROTEE might ben-
efit from a closer focus on mechanisms of institutional
embedding that form a necessary part of successful niche
developments. By limiting its focus to learning pro-
cesses only, the process of technology stabilization and
acceptance might become overlooked. In this respect,
PROTEE can learn from SNM: a focus on processes of
institutional embedding and path dependence may result
in more stable technologies.
We thus conclude that SNM and PROTEE can learn
from one another in achieving a more refined conceptu-
alization of learning and experimenting and in dealing
20 See Hommels (2005). Other mechanisms are the role of dominant
frames, embeddedness and persistent traditions. These mechanisms
play a role at the micro-, meso- and macro-levels of technological
development.
21 This exploration suggests that some of the assumptions underly-
ing the PROTEE approach, unexpectedly and interestingly, are closely
related to recent debates in policy analysis on deliberative governance
(Hajer and Wagenaar, 2003). Considered from a PROTEE perspective,
policy experiments would no longer possess a strategic character which
focuses on the desired end result of a policy decision, but instead would
outline the procedures and principles that could enhance the quality of
the learning process which would lead to that outcome. The outcome
of the process would be the object of debate and criticism. This means
that the end result of a policy experiment can be completely different
from what actors expected it to be at its inception. We believe that our
exploration of these two models could be a next step in the investigation
of the relevance of technology studies for the debate on deliberative
governance.
with the problem of change and obduracy in manag-
ing innovation projects. It will perhaps be difficult to
reconcile the two approaches with respect to the more
fundamental distinction in experimental set-up: the dif-
ference between nurturing innovations in niches and
putting them to the test in a socio techno therapy. This
latter approach makes innovators immediately aware
of the broader context of their innovations instead of
postponing its relevance to a later stage of an innova-
tion’s development. By initially protecting innovations
in niches, as in the SNM approach, there is a risk that
the gap between the innovation and the real world in
which it must be implemented becomes too large. How-
ever, a diminishing of the distinction between niche and
regime as proposed in recent SNM literature (Raven,
2005), opens opportunities to bring to two models closer
to one another: in this way the difference between ini-
tially protecting innovations in niches and making them
vulnerable from the outset becomes smaller.
We think that the development of policy tools like
SNM and PROTEE in technology studies can be fruit-
ful for both STS and the policy arena.22 Our exploration
also shows how the two models, one based on an evo-
lutionary model of technological change and the other
on a constructivist model, can learn from one another.
Apart from bringing the notions of learning and exper-
imenting closer, it would be fruitful to further explore
conceptual connections between notions as path depen-
dence and obduracy and between actor perspectives and
institutional embedding.
Acknowledgements
We would like to thank the participants of the
SNM/PROTEE workshop in Eindhoven in November
2005 and the members of the research group on Science,
Technology and Society of our faculty for their feedback
and critique on an earlier version of this paper. We are
also grateful to Michel Callon and the three anonymous
referees, who provided numerous useful suggestions,
and to Margaret Meredith for editing the English.
References
Berghoff, G., Kneinschnittger, A., Ostkamp, P., 1993. ECT
Delta/Sealand-Terminal Rotterdam, Evaluation of the Technical
and Operational Capacities (evaluation report): Fraunhofer Instit¨
ut
f¨
ur Materialfluß und Logistik.
22 In this sense we agree with Sørensen and Williams (2002) who
argued that technology studies can help to broaden the agenda for
technology policy.
Author's personal copy
A. Hommels et al. / Research Policy 36 (2007) 1088–1099 1099
Bijker, W.E., Hommels, A., Peters, P., 1999. The ECT Delta/Sea-Land
Terminal—a Case of Container Terminal Automation. Universiteit
Maastricht, Maastricht.
Bijker, W.E., 2006. The vulnerability of technological culture. In:
Nowotny, H. (Ed.), Cultures of Technology and the Quest for
Innovation. Berghahn Books, New York.
Brown, H.S., Vergragt, P., Green, K., Berchicci, L., 2003. Learning for
sustainability transition through bounded socio-technical exper-
iments in personal mobility. Technology Analysis and Strategic
Management 15 (3), 291–315.
Garud, R., Karnøe, P. (Eds.), 2001. Path Dependence and Creation.
Lawrence Erlbaum Associates Publishers, Mahwah.
Geels, F.W., 2002. Towards sociotechnical scenarios and reflexive
anticipation: using patterns and regularities in technology dynam-
ics. In: Sørensen, K.H., Williams, R. (Eds.), Shaping Technology,
Guiding Policy, Concepts, Spaces and Tools. Edward Elgar Pub-
lishing, Cheltenham, pp. 359–385.
Geels, F.W., 2004. From sectoral systems of innovation to socio-
technical systems. Insights about dynamics and change from
sociology and institutional theory. Research Policy 33, 897–920.
Hajer, M., Wagenaar, H. (Eds.), 2003. Deliberative Policy Analysis.
Understanding Governance in the Network Society. Cambridge
University Press, Cambridge.
Hommels, A., 2005. Unbuilding Cities. Obduracy in Urban Sociotech-
nical Change. MIT Press, Cambridge, MA.
Hoogma, R., Kemp, R., Schot, J., Truffer, B., 2002. Experimenting for
sustainable transport. The approach of Strategic Niche Manage-
ment. Spon Press, London.
Jolivet, E., Laredo, P., Shove, E. (2003), Managing breakthrough inno-
vations: theoretical implications from – and for – the sociology of
science and technology in Paper presented at the ASET Confer-
ence, Manchester.
Kemp, R., Rip, A., Schot, J., 2001. Constructing transition paths
through the management of niches. In: Garud, R., Karnøe, P. (Eds.),
Path Dependence and Creation. Lawrence Erlbaum Associates
Publishers, Mahwah.
Kets, A., Mourik, R., 2003. Inbedding van innovaties. Over de
waarschijnlijkheid en wenselijkheid van toekomstbeelden. Arena
9, 148–150 (December).
Latour, B., 1997. in: Hofstede, R., Kooy, H.v.d. (Trans.), De Berlijnse
Sleutel en andere lessen van een liefhebber van wetenschap en
techniek, Amsterdam: Van Gennep.
MacKenzie, D., Wajcman, J., 1999. Introductory essay: the social shap-
ing of technology. In: MacKenzie, D., Wajcman, J. (Eds.), The
Social Shaping of Technology. Open University Press, Bucking-
ham, pp. 3–27.
Misa, T.J., 1988. How machines make history, and how historians (and
others) help them to do so. Science, Technology, and Human Values
13 (3–4), 308–331.
Peters, P., 2006. Time, Innovation and Mobilities. Travel in Techno-
logical Cultures. Routledge, London.
Pinch, T., 2001. Why do you go to a piano store to buy a synthe-
sizer: path dependence and the social construction of technology.
In: Garud, R., Karnøe, P. (Eds.), Path Dependence and Creation.
Lawrence Erlbaum Associates Publishers, Mahwah.
PROTEE final report (EU 4th framework) (1999, October).
Retrieved December 15, 2004, from http://europa.eu.int/comm/
transport/extra/final reports/strategic/PROTEE.pdf.
Raven, R. (2005) Strategic Niche Management for Biomass. A com-
parative study on the experimental introduction of bioenergy
technologies in the Netherlands and Denmark PhD Thesis Tech-
nische Universiteit Eindhoven, Eindhoven.
Rip, A., Kemp, R., 1998. Technological change. In: Rayner, S., Malone,
E.L. (Eds.), Human Choice and Climate Change (vol. II Resources
and technology). Batelle Press, Columbus, pp. 327–399.
Rip, A., Misa, T.J., Schot, J. (Eds.), 1995. The Approach of Construc-
tive Technology Assessment. Pinter Publishers, New York.
Russell, S., Williams, R., 2002. Social shaping of technology: frame-
works, findings and implications for policy with glossary of social
shaping concepts. In: Sørensen, K.H., Williams, R. (Eds.), Shaping
Technology, Guiding Policy, Concepts, Spaces and Tools. Edward
Elgar Publishing, Cheltenham, pp. 37–131.
Schumpeter, J., 1942. Capitalism, Socialism and Democracy. Harper
& Row, New York.
Smith, A., Stirling, A., Berkhout, F., 2005. The governance of sus-
tainable socio-technical transitions. Research Policy, 1491–1510,
34.
Sørensen, K.H., Williams, R. (Eds.), 2002. Shaping Technology, Guid-
ing Policy, Concepts, Spaces and Tools. Edward Elgar Publishing,
Cheltenham.
Weber, M., Hoogma, R., Lane, B., Schot, J., 1999. Experimenting
with Sustainable Transport Innovations. A Workbook for Strategic
Niche Management. Promotioneel Drukwerk Service, Enschede,
Seville.